Face Recognition Using Curvelet Transform

This paper presents a new method for the problem of human face recognition from still images. This is based on a multiresolution analysis tool called Digital Curvelet Transform. Curvelet transform has better directional and edge representation abilities than wavelets. Due to these attractive attributes of curvelets, we introduce this idea for feature extraction by applying the curvelet transform of face images twice. The curvelet coefficients create a representative feature set for classification. These coefficients set are then used to train gradient descent backpropagation neural network (NN). A comparative study with wavelet-based, curvelet-based, and traditional Principal Component Analysis (PCA) techniques is also presented. High accuracy rate of 97% and 100% achieved by the proposed method for two well-known databases indicates the potential of this curvelet based curvelet feature extraction method.